Association Between Hereditary Lobular Breast Cancer Due to CDH1 Variants and Gastric Cancer Risk

Key Points

Question  Is there an association between hereditary lobular breast cancer due to CDH1 variants and gastric cancer risk?

Findings  In this cohort study of 283 patients with hereditary lobular breast cancer due to CDH1 pathogenic or likely pathogenic variants, a high prevalence of occult signet ring cell gastric cancer was found.

Meaning  Gastric cancer risk assessment and counseling for patients with hereditary lobular breast cancer should be informed by germline CDH1 variants irrespective of family history of cancer.

Importance  Hereditary cancer risk is informed by the presence of a germline gene variant more so than by family history of cancer.

Objective  To assess gastric cancer risk among patients who received a diagnosis of hereditary lobular breast cancer (HLBC) owing to a germline loss-of-function variant in CDH1 by establishing prevalence of signet ring cell carcinomas among asymptomatic patients.

Design, Setting, and Participants  A prospective cohort study of patients with germline CDH1 pathogenic or likely pathogenic (P/LP) variants at a quaternary medical center were enrolled between October 2017 and January 2021. Data analysis was performed in May 2021. Analyses for associations were performed for these 3 patient groups: (1) family history of breast cancer and no gastric cancer in the HLBC group; (2) family history of gastric cancer and no breast cancer in the hereditary diffuse gastric cancer (HDGC) group; and (3) family history of both breast and gastric cancers in the mixed group. Categorical variables were compared using the Pearson χ² test.

Main Outcomes and Measures  The primary end point of this study was the prevalence of occult signet ring cell carcinoma of the stomach in patients with HLBC. Personal and family medical history, genotype, and pathologic data from risk-reducing total gastrectomy and surveillance endoscopy were examined.

Results  A total of 283 patients with CDH1 P/LP variants (199 [70.3%] were female, and 259 [91.5%] were White; median age, 48 years [range, 18-81 years]) were enrolled in a prospective study of HDGC. The cohort consisted of 151 families. Patients were categorized according to family history of breast and/or gastric cancer: HLBC 15.5% [44 of 283 patients]), HDGC (16.2% [46 of 283 patients]), and mixed (68.2% [193 of 283 patients]). The HLBC group included 31 distinct families with 19 CDH1 variants; 10 of those variants were also present in the HDGC and mixed groups (52.6% [10 of 19 variants]). Nearly all of the patients with HLBC (93.8% [15 of 16 variants]) who elected for risk-reducing total gastrectomy owing to their underlying CDH1 P/LP variant harbored occult signet ring cell gastric adenocarcinoma on final pathology (median age, 50 years [range, 21-67 years]). The prevalence of occult gastric cancer among asymptomatic patients in the HDGC group was similar (94.7% [18 of 19 of variants]; P = .98).

Conclusions and Relevance  Carriers of CDH1 P/LP variants with no family history of gastric cancer exhibited high rates of occult signet ring cell gastric cancer. Germline CDH1 P/LP variants appear to have a highly penetrant gastric phenotype irrespective of family history. These data may prove useful for counseling families with CDH1 variants presumed to have HLBC.

Hereditary lobular breast cancer (HLBC) has been proposed as an independent cancer syndrome in families with germline CDH1 (OMIM 192090) pathogenic or likely pathogenic (P/LP) variants without a family history of gastric cancer.^1,2 Loss-of-function variants in the CDH1 tumor suppressor gene are commonly known to be result in the hereditary diffuse gastric cancer (HDGC) syndrome, which includes an elevated risk of lobular breast cancer.^3,4 Contemporary penetrance analyses estimate that, among patients with HDGC, the lifetime risk of gastric cancer is 33% to 42% and the lifetime risk of breast cancer is 43% to 55%.^5,6 Because of the aggressive biology of and lack of effective surveillance for HDGC, risk-reducing total gastrectomy is recommended for CDH1 variant carriers.^2,5,6 The risk of gastric cancer in patients with HLBC is unknown, even though occult signet ring cell carcinoma is a frequent finding in patients with HDGC.⁷ Thus, the potential for a pure HLBC phenotype may affect genetic cancer risk assessment and counseling. We hypothesized that family history of breast cancer is an independent factor with no association with gastric cancer phenotype in CDH1 variant carriers. We sought to assess the prevalence of occult gastric cancer among patients with HLBC and determine genotype-phenotype correlations in a cohort of CDH1 P/LP variant carriers.

CDH1 variant carriers enrolled in a prospective cohort study of hereditary gastric cancer between October 2017 and January 2021 were examined. Written informed consent was obtained from all participants at the time of study enrollment, and approval by the institutional review board of the National Cancer Institute, National Institutes of Health was obtained. Prospective data included patient demographic characteristics, CDH1 genotype, family pedigrees, and gastric and breast pathological findings. Previous gastric and breast pathological findings were confirmed when possible. Race and ethnicity and sex were self-reported by participants; race and ethnicity were assessed for external validity. Genotype analysis included variant type and E-cadherin domain affected. History of gastric cancer was recorded for clinically diagnosed (generally advanced) cancers. Diagnosis of occult cancer after risk-reducing total gastrectomy was not considered positive family history of gastric cancer. Patients were categorized according to family history of breast and gastric cancers. Analyses for associations were performed for these 3 patient groups: (1) family history of breast cancer and no gastric cancer in the HLBC group; (2) family history of gastric cancer and no breast cancer in the HDGC group; and (3) family history of both breast and gastric cancers in the mixed group.

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Categorical variables were compared using the Pearson χ² test. Variables with P < .05 were considered statistically significant. Multinomial logistic regression analysis was not possible given the small patient groups. Data were analyzed with SPSS for Windows, version 25.0 (SPSS Inc).

Overall, 283 patients with CDH1 P/LP variants were analyzed (Table 1). Most patients were female (199 [70.3%]) and White (259 [91.5%]), with median age of 48 years (range, 18-81 years) at enrollment. A personal history of breast cancer was present for 58 patients (20.4%) diagnosed at median age of 49 years (range, 32-78 years), whereas gastric cancer was diagnosed in 7 patients (2%) at median age of 40 years (range, 28-60 years). In 151 families, there were 455 relatives who received a diagnosis of breast cancer (median age, 55 years) and 466 relatives who received a diagnosis of gastric cancer (median age, 51 years). Variants were most frequently nonsense variants (32%) and were distributed throughout all CDH1 domains (eTable in the Supplement). Patients were categorized according to family history of breast and gastric cancers: HLBC (15.5% [44 of 283 patients]), HDGC (16.2% [46 of 283 patients]), and mixed (68.2% [193 of 283 patients]). We examined the HLBC group with specific attention to CDH1 genotype and prevalence of occult gastric cancer. This group consisted of 31 families with 19 CDH1 variants. Two of the 44 patients (4.5%) carried an additional pathogenic germline variant (CHEK2 or NTHL1). Sixteen patients (36.4%) had a personal history of breast cancer (median age at diagnosis, 47 years [range, 35-65 years]); family history of breast cancer was present in 90 relatives (median age, 52 years). By definition, there was no personal or familial history of gastric cancer. Nearly half of the HLBC group’s variants were nonsense variants (21 of 44 [47.7%]), and the most frequent location was the cadherin 4 domain (13 [29.5%]). We compared the HLBC group’s genotypes with the genotypes of the entire cohort and found that 10 of 19 CDH1 variants were also present in families with documented cases of gastric cancer. CDH1 genotypes associated with HLBC in our cohort were compared with genotypes previously reported as unique to HLBC (Table 2).^8-13

One or more surveillance endoscopies were performed for 32 of the 44 patients (72.7%) with HLBC. Occult signet ring cell carcinoma was detected in 11 of the 32 patients (34.4%; median age, 33 years [range, 27-79 years]). Nearly all patients with HLBC (15 of 16 [93.8%]) who elected for total gastrectomy harbored gastric adenocarcinoma (median age, 50 years). At the time of prophylactic total gastrectomy, 14 patients had occult carcinoma, and 1 patient had a small ulcer on preoperative endoscopy that proved to be a T3 adenocarcinoma prompting adjuvant chemotherapy. By comparison, 30 patients in the HDGC group underwent surveillance, 43.3% (13 of 30) had occult carcinoma (median age, 43 years), and all but 1 patient (18 of 19 [94.7%]) had occult gastric cancer at the time of gastrectomy (median age, 33 years) (Table 3). We assessed for genotype-phenotype correlations that might explain the variable penetrance of breast and gastric cancers in this cohort. Nonsense variants were more likely to occur in the HLBC and HDGC groups (47.7% [21 of 44] and 41.3% [19 of 46], respectively) compared with the mixed group (26.9% [52 of 193]) (P = .01). Splice site variants were most common in the mixed group (31%; P = .008) and more frequent in the HLBC group compared with the HDGC group (25% vs 8%; P = .04). Genotype domain location did not correlate with phenotype (Figure).

We have shown that occult gastric adenocarcinoma is highly prevalent in HLBC families with CDH1 variants. These data support the claim that the risk of occult gastric cancer remains high for patients with HLBC irrespective of family history. In total, the data presented here expose the limitations of family history when used as criteria for cancer risk assessment and counseling of patients with germline CDH1 variants. We discovered several CDH1 variants purportedly unique to HLBC that were actually present in families with documented gastric cancer.^1,10 Corso et al reported an association between missense variants and HLBC phenotypes.¹ In our cohort, the most common CDH1 variant type in HLBC was nonsense, whereas missense variants occurred less than 10% of the time. Linker region variants have been reported as protective of CDH1-associated breast cancer.⁶ However, our findings demonstrated that these variants were present in all 3 patient groups at similar rates. These contradictory findings are likely due to additional, yet unknown, molecular alterations resulting in complex cancer phenotypes.

Despite a large sample size, this study of a hereditary cancer syndrome is limited by ascertainment bias.² Disease phenotype was established from family pedigrees, which has the potential for recall bias by the proband. To improve the accuracy of each family history, pedigree updates were verified prior to data analysis. The HLBC, HDGC, and mixed patient groups were generated to facilitate comparisons. It is possible that segregation of patients based on the strength or weakness of cancer history using quantitative criteria could have yielded different results. We omitted diagnosis of occult gastric cancer at risk-reducing gastrectomy from family history; the decision to focus on clinically diagnosed cancers was deliberate, even though the rate of progression of occult cancers is not known.

In this cohort study, germline CDH1 P/LP variants were not associated with mutually exclusive HLBC and HDGC phenotypes; furthermore, the rates of occult gastric carcinoma were the same for patients with HLBC and patients with HDGC. Thus, it is paramount that patients previously assigned a diagnosis of HLBC are not excluded from receiving a proper gastric cancer risk assessment and counseling. The mere presence of a germline CDH1 P/LP variant, regardless of family history, may be reason enough to consider prophylactic total gastrectomy. Although it is possible that patients with HLBC may not develop clinically evident gastric cancer, the mere existence of occult carcinomas evokes clinical and scientific questions. Only when factors associated with gastric cancer quiescence and progression are elucidated will the relative risks of gastric and breast cancers in CDH1 variant carriers likely be known.

Accepted for Publication: August 4, 2021.

Published Online: October 13, 2021. doi:10.1001/jamasurg.2021.5118

Corresponding Author: Jeremy L. Davis, MD, Surgical Oncology Program, Center for Cancer Research, National Cancer Institute, National Institutes of Health, 10 Center Dr, Room 4-3742, Bethesda, MD 20892 (jeremy.davis@nih.gov).

Author Contributions: Dr Davis had full access to all the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis.

Acquisition, analysis, or interpretation of data: Gamble, Rossi, Fasaye, Kesserwan, Hernandez, Blakely.

Drafting of the manuscript: Gamble, Rossi, Hernandez, Davis.

Critical revision of the manuscript for important intellectual content: Rossi, Fasaye, Kesserwan, Blakely, Davis.

Statistical analysis: Gamble, Rossi, Blakely.

Administrative, technical, or material support: Fasaye, Kesserwan.

Supervision: Blakely, Davis.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported in part by the Intramural Research Program of the National Cancer Institute, National Institutes of Health.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We wish to thank the patients and families who helped make this research possible.